1. Field of the Invention
The present invention relates generally to risk analysis systems, and, in particular, to computer-based risk assessment systems.
2. Description of the Related Art
The prior art includes many software tools, primarily CAFTA, ETA, RISKMAN, and IRRAS. In addition, performing risk analysis through the construction of fault trees is known.
The state-of-the-art in risk assessment software is now described. The current software packages/systems for probabilistic risk assessment (PRA) utilize a fault-tree model approach, an event tree model approach, or event trees with fault trees attached to branch points. Two main examples are CAFTA for WINDOWS and Riskman. Other software of the same genre includes IRRAS--with the WINDOWS based version called SAPPHIRE.
The fundamental problem with fault trees, though, is they only provide an upper bound to the true risk (and this is not a least upper bound). Riskman and CAFTA use fault trees. CAFTA is essentially a large fault tree program. Riskman uses top level event trees and, in general, hangs fault trees at the branch points.
Therefore, the Quantitative Risk Assessment System (QRAS) model of the present invention, as explained herein below, gives a better approximation to the true but unknown risk.
From a structural point of view, a fault tree only looks at the base events and creates cut sets, and then finds the minimal cut sets. However, none of the prior art software handles all of the elements simultaneously. The prior art software does not allow one to change an element (a failure mode or a failure mode quantification) and have it changed everywhere it applies. If one changes the set up of a system in CAFTA, the prior generated cut sets still exist (in file form).
On the other hand, in QRAS of the present invention, as explained herein below, if a user changes the structure (e.g., a failure mode or a failure mode quantification) of the system for which risk is being analyzed, then the baseline will be removed and all analysis runs pertaining to it will not exist. (More particularly, in QRAS of the present invention, the user must first supply a password to delete the baseline. Alternatively in QRAS of the present invention, the entire project can be saved and a new project created by a copy/paste function or simply by using WINDOWS Explorer, and then this new project can be modified.) That is, in QRAS of the present invention, all of the analysis runs will not exist unless the entire model is preserved unchanged. In CAFTA, on the other hand, a cut set file can exist, even though one can then change the original fault tree model and the cut set file, although inaccurate, refers both to the old system (because it has the refer-back name), but it does not really apply because the system has changed.
In addition, event sequence diagrams are known. Also known are failure modes, as are demand-based and time-based quantifications of failure modes, such as point estimate/uncertainty distribution, writing the probability (P) as a function of variables, logistic regression for P, limit state functions, exponential with uncertainty in .lambda., piecewise exponential, weibull--with uncertainty on .beta., .eta., standby redundancy, conditional probability (exponential), and conditional probability (weibull).
Also known is risk ranking individually by mean, by median, and by uncertainty (for a scenario or a failure mode). No software makes a distinction between scenarios that immediately propagate to failure (i.e., single point failures) and scenarios with one or more pivotal events, as in the present invention. No software internally collects the probabilities of failure over scenarios and then ranks the individual elements or subsystems using those fundamental units of failure, as in the present invention. A singleton, which is an initiating event followed immediately by an end state, itself is known, but the present invention allows an immediate determination of what are those single point failures (i.e., singletons).
There are, in addition, WINDOWS-based software programs for creating and analyzing reliability block diagrams, and for incorporating uncertainties. There is also WINDOWS-based software for failure modes and effects analysis, etc.
However, there is no conventional risk assessment software incorporating as a unit the following: ease-of-use; the availability of as many quantification methods; the linkage via a hierarchy (with plain text for names) to a time-based quantification and the abilities to store documentation; and perform sensitivity analysis (linked to original risk analysis)--as in the present invention.
Although each individual quantification or statistical method used by QRAS of the present invention is known, the combination of these methods used as they are in QRAS of the present invention is unique. Moreover, there is no other software that handles event sequence diagrams (ESDs) and automatically translates the ESDs into event trees, as in the present invention.
Further, the prior art systems are not particularly user-friendly. More specifically, none of the prior art systems allows the range of failure probability characterizations as in the present invention, none are based on a hierarchical arrangement with the features as in the present invention, none include a WINDOWS-based event-sequence diagram builder to automatically create event trees, as in the present invention. In addition, the prior art systems include limited sensitivity analysis capabilities. Also, and most importantly, the prior art systems do not integrate the above-mentioned features in a cohesive, simple, yet powerful platform.